Moonshot AI Kimi K2-Thinking-0905: LLM Security & Jailbreak Resistance
As of 2026-07-04, Moonshot AI Kimi K2-Thinking-0905 has an attack success rate (ASR) of 38.0% on Guardion's LLM Vulnerability benchmark — ranking #50 of 93 models. It is weakly resistant — adversarial prompts succeed often without added guardrails.
Attack Success Rate (ASR) is the share of adversarial prompts that elicit harmful output across zero-shot, TAP, and Crescendo attacks (HarmBench framework). Lower is safer. "Est." models are calibrated from public safety evaluations pending a Guardion benchmark run.
How Kimi K2-Thinking-0905 compares
Frequently asked questions
How secure is Moonshot AI Kimi K2-Thinking-0905 against jailbreaks?
Moonshot AI Kimi K2-Thinking-0905 is weakly resistant — adversarial prompts succeed often without added guardrails. On Guardion's LLM Vulnerability benchmark it records a 38.0% attack success rate (62% robustness), ranking #50 of 93 models (estimated from public safety evaluations).
What is Kimi K2-Thinking-0905's attack success rate (ASR)?
Kimi K2-Thinking-0905's overall ASR is 38.0% — the share of adversarial prompts that elicit harmful output across zero-shot, TAP, and Crescendo attacks. Lower is safer.
Is Kimi K2-Thinking-0905 more secure than Qwen3-Next-80B-A3B-Instruct?
Qwen3-Next-80B-A3B-Instruct is more robust: 37.0% ASR vs Kimi K2-Thinking-0905's 38.0%. See the full head-to-head comparison for the per-attack breakdown.
How can I make Kimi K2-Thinking-0905 safer to deploy?
Even robust models are bypassed under sustained attack. An inline runtime guardrail that classifies prompts and responses — like Guardion's prompt-defense models — blocks jailbreaks and prompt injection before they reach Kimi K2-Thinking-0905.
Harden Kimi K2-Thinking-0905 in production
Guardion's runtime guardrails block jailbreaks and prompt injection before they reach any model — with sub-130ms policy decisions.
Request a demo